Sorry, you need to enable JavaScript to visit this website.

Non-parametric Bayesian methods have recently gained popularity in several research areas dealing with unsupervised learning. These models are capable of simultaneously learning the cluster models as well as their number based on properties of a dataset. The most commonly applied models are using Dirichlet process priors and Gaussian models, called as Dirichlet process Gaussian mixture models (DPGMMs). Recently, von Mises-Fisher mixture models (VMMs) have also been gaining popularity in modelling high-dimensional unit-normalized features such as text documents and gene expression data.

This paper deals with the separation of music into individual instrument tracks which is known to be a challenging problem. We describe two different deep neural network architectures for this task, a feed-forward and a recurrent one, and show that each of them yields themselves state-of-the art results on the SiSEC DSD100 dataset. For the recurrent network, we use data augmentation during training and show that even simple separation networks are prone to overfitting if no data augmentation is used.

The paper introduces a novel class of complex nonlinear filters, the complex functional link polynomial (CFLiP) filters.
These filters present many interesting properties. They are a sub-class of linear-in-the-parameter nonlinear filters.
They satisfy all the conditions of Stone-Weirstrass theorem and thus are universal approximators for causal, time-invariant, discrete-time, finite-memory, complex, continuous systems defined on a compact domain.

First-person action recognition is a recent problem in computer vision, where an observer wears body cameras to understand and recognize actions from the captured video sequences. Technological advances have made it possible to offer small wearable cameras that can be attached onto bike helmets, belts, animal halters, among other accessories. Examples of potential applications include sports, security, healthcare, visual lifelogging, among others.

This paper studies optimal transmit strategies for multiple-input multiple-output (MIMO) Gaussian channels with joint sum and per-antenna power constraints. It is shown that if an unconstraint optimal allocation for an antenna exceeds a per-antenna power constraint, then the maximal power for this antenna is used in the constraint optimal transmit strategy. This observation is then used in an iterative algorithm to compute the optimal transmit strategy in closed-form. Finally, a numerical example is provided to illustrate the theoretical results.

It is well known that the positive- and negative frequency components of a real sinusoid spectrally interact with each other; thus, introducing bias in frequency estimation based on the periodogram maximization. We propose to filter out the negative-frequency component. To that end, a coarse frequency estimation is obtained using the windowing approach, known to reduce the estimation bias, and then used to filter out the negative frequency component via modulation and discrete Fourier transform

Cellular backhaul networks usually consist of commercial microwave links, known to be sensitive to weather conditions. The management network systems usually provide records of measurements of the transmitted and the received signals levels from the different microwave links for monitoring and analyzing the network performance. Many of them log only the minimum and the maximum levels of the transmitted and the received signals in pre-set intervals (usually 15-minute). Moreover, only quantized version of these measurements are logged.

We propose a novel application of the acoustic- to- articulatory inversion (AAI) towards a quality assessment of the voice converted speech. The ability of humans to speak effortlessly requires the coordinated movements of various articulators, muscles, etc. This effortless movement contributes towards a naturalness, intelligibility and speaker’s identity (which is partially present in voice converted speech). Hence, during voice conversion (VC), the information related to the speech production is lost.

A new approach to online Speech Activity Detection (SAD) is proposed. This approach is designed for the use in a system that carries out 24/7 transcription of radio/TV broadcasts containing a large amount of non-speech segments. To improve the robustness of detection, we adopt Deep Neural Networks (DNNs) trained on artificially created mixtures of speech and non-speech signals at desired levels of Signal-to-Noise Ratio (SNR). An integral part of our approach is an online decoder based on Weighted Finite State Transducers (WFSTs); this decoder smooths the output from DNN.